Martingales #
THIS FILE IS SYNCHRONIZED WITH MATHLIB4. Any changes to this file require a corresponding PR to mathlib4.
A family of functions f : ι → Ω → E
is a martingale with respect to a filtration ℱ
if every
f i
is integrable, f
is adapted with respect to ℱ
and for all i ≤ j
,
μ[f j | ℱ i] =ᵐ[μ] f i
. On the other hand, f : ι → Ω → E
is said to be a supermartingale
with respect to the filtration ℱ
if f i
is integrable, f
is adapted with resepct to ℱ
and for all i ≤ j
, μ[f j | ℱ i] ≤ᵐ[μ] f i
. Finally, f : ι → Ω → E
is said to be a
submartingale with respect to the filtration ℱ
if f i
is integrable, f
is adapted with
resepct to ℱ
and for all i ≤ j
, f i ≤ᵐ[μ] μ[f j | ℱ i]
.
The definitions of filtration and adapted can be found in probability.stopping
.
Definitions #
measure_theory.martingale f ℱ μ
:f
is a martingale with respect to filtrationℱ
and measureμ
.measure_theory.supermartingale f ℱ μ
:f
is a supermartingale with respect to filtrationℱ
and measureμ
.measure_theory.submartingale f ℱ μ
:f
is a submartingale with respect to filtrationℱ
and measureμ
.
Results #
measure_theory.martingale_condexp f ℱ μ
: the sequenceλ i, μ[f | ℱ i, ℱ.le i])
is a martingale with respect toℱ
andμ
.
A family of functions f : ι → Ω → E
is a martingale with respect to a filtration ℱ
if f
is adapted with respect to ℱ
and for all i ≤ j
, μ[f j | ℱ i] =ᵐ[μ] f i
.
Equations
- measure_theory.martingale f ℱ μ = (measure_theory.adapted ℱ f ∧ ∀ (i j : ι), i ≤ j → measure_theory.condexp (⇑ℱ i) μ (f j) =ᵐ[μ] f i)
A family of integrable functions f : ι → Ω → E
is a supermartingale with respect to a
filtration ℱ
if f
is adapted with respect to ℱ
and for all i ≤ j
,
μ[f j | ℱ.le i] ≤ᵐ[μ] f i
.
Equations
- measure_theory.supermartingale f ℱ μ = (measure_theory.adapted ℱ f ∧ (∀ (i j : ι), i ≤ j → measure_theory.condexp (⇑ℱ i) μ (f j) ≤ᵐ[μ] f i) ∧ ∀ (i : ι), measure_theory.integrable (f i) μ)
A family of integrable functions f : ι → Ω → E
is a submartingale with respect to a
filtration ℱ
if f
is adapted with respect to ℱ
and for all i ≤ j
,
f i ≤ᵐ[μ] μ[f j | ℱ.le i]
.
Equations
- measure_theory.submartingale f ℱ μ = (measure_theory.adapted ℱ f ∧ (∀ (i j : ι), i ≤ j → f i ≤ᵐ[μ] measure_theory.condexp (⇑ℱ i) μ (f j)) ∧ ∀ (i : ι), measure_theory.integrable (f i) μ)
The converse of this lemma is measure_theory.submartingale_of_set_integral_le
.
A predictable submartingale is a.e. greater equal than its initial state.
A predictable supermartingale is a.e. less equal than its initial state.
A predictable martingale is a.e. equal to its initial state.
Given a discrete submartingale f
and a predictable process ξ
(i.e. ξ (n + 1)
is adapted)
the process defined by λ n, ∑ k in finset.range n, ξ (k + 1) * (f (k + 1) - f k)
is also a
submartingale.